Advanced Computer Architecture - Lecture 43: Networks and clusters. This lecture will cover the following: internetworks; cluster; case studies; OSI layers; Transmission Control Protocol/Internet Protocol (TCP/IP); non-standard connections; division of memory;...
CS 704 Advanced Computer Architecture Lecture 43 Networks and Clusters (Internetworks and Clusters) Prof Dr M Ashraf Chughtai Today’s Topics Recap: Internetworks Cluster Case Studies Summary MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) Recap: In our last two lectures on Networks and Cluster we discussed: The formation of generic interconnection networks and their categorization, the networks communication model, performance, media, software, protocols, subnet and networks topologies Here, we noticed that a generic interconnection network comprises: Computer nodes, H/W and S/W interface, Links to the interconnection network and Communication subnet MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) Recap The interconnect communication model shows that two machines are connected via two unidirectional wires with a FIFO (queue) at the end to hold the data The communication software separates the header and trailer from the message and identifies the request, reply, their acknowledgments and error checking codes The communication protocols suggest the sequence of steps to reliable communication MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) Recap We also discussed: the properties and performance of interconnect network media or link – the unshielded twisted pair (UTP), coaxial cable and fiber optics the formation of bus-based and switch-based communication subnets and introduced the network topologies The bus-based communication subnets share the common media where arbitration is the bottleneck MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) Recap Alternative to sharing media is to use a switch to provide a dedicated line to all destinations in order; and facilitates point-to-point communication much faster than the shared media The switch-based networks are classified as the centralized and distributed switch networks Here the routing, to establish interconnection between two node at a time, depends on the addressing style: source-based routing and destination-based routing MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) Recap The performance of a distributed network is measured in terms of: – Latency - number of Links between source and destination – Bandwidth – number or length of messages passing per sec – Degree - number of links connected to a node – Diameter - number of nodes between source and destination; this is in deed the measure of maximum latency MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) Recap – Bisection - the imaginary line that divides the interconnect into roughly two equal parts, each having half the nodes – Bisection Bandwidth: the volume of communication allowed between any two halves of network with equal number of nodes Last time, we discussed an intermediate class of network interconnect – Multistage Switch network It is built from number of large switch boxes each containing number of small crossbar MAC/VU-Advanced switches Computer Architecture Lecture 43 Networks and Clusters (3) Recap The performance of Multi-stage switch lies between performance of non-locking crossbar and bus-based networks Following the discussion on centralized switch topologies we studied the distributed-switch interconnects; which are categorized as the fully-connected and partially-connected, symmetric or asymmetric interconnects The distributed-switch interconnect topologies, such as: linear array, ring, 2D mesh/torus and hypercube were studied MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) Recap We also discussed the relative cost and performance of these topologies, based on the bisection bandwidth and number of links for 64 nodes network; which is shown in the following table Evaluation Category Bus Ring 2D Torus Fully Connected Performance: Bisection B/W 16 1024 64 128 192 Cost -Ports/switch MAC/VU-Advanced Computer Architecture N/A Total Links 2080 Lecture 43 Networks and Clusters (3) 10 Cluster Design Examples … Cont’d Example Now let us discuss the 4th case, cluster design for transaction processing, shown here Fig 8.40 pp 853 The cluster has 32 P-III processors, using the same IBM computer as the basic building block and the switch as was employed in earlier designs MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 42 Cluster Design Examples … Cont’d The key differences are: Disk Size: Small and fast disks are used as this structure cares more about I/Os per second (IOPS) RAID: No RAID is required as this as the performance benchmark doesn't include the human cost Memory: Maximum DRAM is packed into the servers, so each of the four 8-way SMPs is stuffed with maximum of 32 GB, yielding 128 GB MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 43 Cluster Design Examples … Cont’d Processor: 900 MHz P-III with MB L2 cache is used The cost performance analysis of this structure shows that almost half of the cost is in software, installation and maintenance Summarizing the cluster design examples we conclude that as the cost of purchase is less than half the cost of the ownership, therefore, the cost of hardware solves only a part of the problem MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 44 Summary In this module on the Networks and clusters we studied that: The formation of a generic interconnection network that comprises – – – – Computer nodes (host or end system) H/W and S/W interface Links to the interconnection network and Communication subnet The interconnections are designated as: – Local Area Network-LAN – Wide Area Network-WAN – System (or Storage) Area Network-SAN MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 45 Summary While talking about the interconnect model, software and protocols we studied that: – The interconnect communication model shows that two machines are connected via two unidirectional wires with a FIFO (queue) at the end to hold the data – The communication software separates the header and trailer from the message and identifies the request, reply, their acknowledgments and error checking codes – The communication protocols suggest the sequence of steps to reliable communication MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 46 Summary Then we studied the network performance that defines the latency of the message as the sum of Sender overhead, time to flight, receiver overhead and the ratio of the message size to the bandwidth We also discussed the properties and performance of interconnect network media or link – the unshielded twisted pair (UTP), coaxial cable and fiber optics; and the formation of bus-based and switch-based communication subnets and introduced the network topologies MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 47 Summary We also studied an intermediate class of network interconnect, which lies between crossbar and bus-based networks, referred to as the Multistage Switch that is built from number of large switch boxes each containing small crossbar switches; here The number of identical stages (Ns) of large switch boxes each having m x m crossbar switches, in the network having n nodes, is equal to log m n; and, the switches per stage is n/m MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 48 Summary The cost of multistage switch network is of O(n log n) which is considerable small as compared to that of crossbar network that is of O(n2 ) when n is large The typical examples of multistage switching topologies discussed are Omega and Butterfly networks Following the discussion on centralized switch topologies we considered the distributedswitch interconnects where the switches are distributed among the nodes (processor) MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 49 Summary We classified the distributed switch interconnects as the fully and partially connected, symmetric and asymmetric interconnects Then we discussed the linear array, ring, 2D mesh/torus and hypercube topologies and their performance measures The relative cost and performance of these topologies, based on the bisection bandwidth and number of links for 64 nodes network is as follows MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 50 Comparing Network Topologies Evaluation Category Performance: Bisection Bandwidth Bus Ring 2D Torus Fully Connected 16 1024 -Ports/switch N/A 64 - Total Links 2080 128 192 Cost MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 51 Summary Today we discussed Internetworking, i.e., the connection of two or more interconnection networks to communicate reliably and efficiently Internetworking relies on the communication standards composed of hierarchy of layers The internet communication Protocol Families facilitates applications to work with any interconnection MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 52 Summary The Transmission Control Protocol/Internet Protocol - TCP/IP is the most popular internetworking standard The protocol at each level is implemented by adding headers and trailers at the sending layer and removing at the receiving layer Have introduced the basic concept of internetworking we discussed the computer cluster which is coordinated use of interconnected computers in a machine room MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 53 Summary Here, we studied Non-standard connections and Division of memory as the performance confront of clusters; furthermore Contrary to these challenges, clusters have advantages in respect of dependability and scalability At the end we studied the practical aspects of cluster designs through four examples MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 54 Conclusion Today we have completed our discussion on almost all topics related to this course on Advanced Computer Architecture In the following last two lecture we will review the complete course following some case studies MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 55 Thanks and Allah Hafiz MAC/VU-Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 56 ... Internetworks may consult literature and books on Computer Networks and Internetworking MAC/VU -Advanced Computer Architecture Lecture 43 Networks and Clusters (3) 21 Clusters – System Area Networks. .. and fiber optics; and the formation of bus-based and switch-based communication subnets and introduced the network topologies MAC/VU -Advanced Computer Architecture Lecture 43 Networks and Clusters. .. MAC/VU -Advanced switches Computer Architecture Lecture 43 Networks and Clusters (3) Recap The performance of Multi-stage switch lies between performance of non-locking crossbar and bus-based networks